Stereospecific homoallylic ring expansions and contractions - Journal

Jian-Jung Pan , Gurusankar Ramamoorthy , and C. Dale Poulter. Organic ... Jonathan B. Johnston, Joel R. Walker, Steven C. Rothman, and C. Dale Poulter...
1 downloads 0 Views 866KB Size
4282

Stereospecific Homoallylic Ring Expansions and Contractions' C. Dale Poulterzasband S. Winstein2' Contribution No. 2453 from the Department of Chemistry, University of California, Los Angeles, California 90024. Received September 2, 1969 Abstract: The preparations and solvolyses of syn- and anti-bicyclo[7.l.0]dec-2-ylp-nitrobenzoate (syn- and anti-1COPNB) are described. In 80% acetone-water, syn-16-OPNB gave hydrocarbons (1273, syn-ldOH (26%), cis-cyclodecen-4-01(cis-18-OH) (4573, and cis-18-OPNB(17Z). The anti-p-nitrobenzoate gave anti-16-OH (71%), trans-18-OH(15%), trans-18-OPNB (13%),and trans-bicyclo[6.2.O]decan-trans-9-ol (trans,trans-19-OH)(1%). Hydrolysis of cis-18-OBs produced hydrocarbons (25Z), syn-1QOH (30%), and cis-18-OH (4573, while similar treatment of trans-18-OBs gave anti-16-OH (85z),trans-18-OH (14%), and trans,trans-19-OH (1%). All of the solvolyses are greater than 99.7Z stereoselective. No crossover between syn and anti reaction pathways was observed. The bicyclo[7.1.0]decane carbon framework is the first in the homologous [n.1.0] series which permits an entire set of homoallylic ring expansions and contractions. The stereospecificity is explained in terms of two isomeric, noninterconverting nonclassical homoallylic ions which are generated stereospecifically from syn-cis or anti-trans precursors.

S

tereochemical studies are often invaluable probes for providing information about the nature of reactive intermediates. Cyclopropylcarbinyl to homoallylic isomerizations have been found to be highly stereoselective in several systems. These systems can be divided into two basically different groups. In 1961, Julia and coworkersad reported that isomerizations of 1 to 2 (RI = alkyl, RZ = R3 = H) were 90-95% stereo-

1

2

selective. Johnson and desired to use the method for stereoselective synthesis of trisubstituted double bonds and found when RI = alkyl, Rz = H, and R3 = CH3, that formation of 2 was 97 stereoselective. The stereoselectivity found for 1 can be explained in terms of current concepts about cyclopropylcarbinyl cations.aa Either 3 or 5 permits participation of the

cyclopropane ring during ionization. Since R1(alkyl) is much larger than R2 (H) conformer 3 predominates, and 4 is formed in preference to 6. Once the cyclopropylcarbinyl cation is generated, the relative stereo~,~ chemistry among R1,Rz,and R3 is f i ~ e d . ~ Homoallylic product 2 derived from nucleophilic attack at the rear of the cyclopropane ring is the major homoallylic isomer. In this system ionization of the preferred conformer is the basis for stereoselectivity found in the homoallylic product. Winstein and his c ~ w o r k e r sdiscovered ~~~~ stereospecific cyclopropylcarbinyl to homoallylic isomerizations in a different system, illustrated by the conversion of 7 to 8 and 9 to 10. It was proposed that ionization dilute 10

OH 7

8 dilute HCIO,

n +

"H 9

4

+OH2

-

R3&9R, R2 5

#.I

R2 U

6

(1) This research was supported in part by the National Science Foundation. (2) (a) This investigation was supported in part by National Institutes of Health Postdoctoral Fellowships l-FZ-GM-29,317-01 and 2-F2-GM-29,3 17-02 from the Institute of General Medical Sciences; (b) address correspondence to: The Department of Chemistry, University of Utah, Salt Lake City Utah 84112; (c) deceased, November 23, 1969. (3) (a) C. D. Poulter, E. C. Friedrich, and S. Winstein, J. Amer. Chem. SOC.,92, 4274 (1970). (b) M. GAsiC, D. Whalen, B. Johnson, and S. Winstein, ibid., 89, 6382 (1967); (c) D. Whalen, M. GaSiC, B. Johnson, H . Jones, and S. Winstein, ibid., 89, 6384 (1967); (d) M. Julia, S. Julia, and S.-Y. Tchen, Bull. SOC.Chim. France, 1849 (1961); (e) S. F. Brady, M. A. Ikon, and W. S.Johnson, J. Amer. Chem. SOC.,90, 2882 (1968).

nu

wn

10

gave a homoallylic cation in which positive charge was predominately delocalized to the cyclopropane carbon bearing the allylic substitutent. No products derived from the other, less substituted homoallylic ion were seen. In these systems, the relative orientation of the cyclopropane ring and the leaving group determined the geometry of the homoallylic cation and eventual homoallylic product. Models suggest that sometimes the molecules even twist into unfaaorable conformations to utilize the more substituted arm of the cyclopropane ring during ionization. If the intermediate in these interconversions is a homoallylic cation, then the reverse cyclizations of 8 to 7 and 10 to 9 should also be stereospecific. This prediction was partially verified by the observation that cis-cyclononen4-yl brosylate gave only syn-cyclopropylcarbinyl prodU C ~ . ~ ' We now describe work with a system in which a

Journal of the American Chemical Society J 92:14 J July 15, 1970

(4) P. von R. Schleyer and V. Buss, ibid., 91, 5880 (1969).

4283 full set of stereospecific homoallylic ring expansions and contractions is observed.

Results The synthetic sequence used to prepare syn-bicyclo[7.1 .O]decan-2-01 (syn-16-OH) and anti-bicyclo[7.1 .O]decan-2-01 (anti-ldOH) is outlined in Scheme I. The Scheme I

11

13

12

98%

J. 0

II

Dilute perchloric acid in 80 dioxane-water isomerized syn-16-OH to cis-cyclodecen-4-01 (cis-18-OH) and a mixture of hydrocarbon^.^ The stereochemistry of the double bond was assigned on the basis of ir spectra. Alcohol cis-lS-OH had strong absorptions at 710 and 725 cm-l and no strong bands between 900 and 990 cm-l. On the other hand, trans-18-OH (vide infra) had a strong absorption at 980 cm-l. The ir spectra of these isomers can be compared with those of ciscyclodecene (781, 763, and 706 cm-') and trans-cyclodecene (972 cm-1).8 An nmr spectrum of cis-18-OH was consistent with the assigned structure. The chemical-shift difference between the protons at Cl and Cz was not large enough at 100 MHz to permit a firstorder analysis of the coupling pattern. Treatment of anti-16-OH with dilute perchloric acid at 75" gave two isomeric products (Scheme II), which Scheme II

& doc" OH

4m;

lo

I

H

-OH

H

15

HCI04

8% dioxane-water+

14

50'

k

~nti-160H

1

Jones oxidation 9%

hydrocarbons

+

18%

P OH

H 17

HClO,

8W dioxane-water

addition of methylene to 15 is highly stereoselective, giving only 0.05% of the syn epimere5 Oxidation of anti-16-OH to 17, followed by reduction with lithium aluminum hydride, gave syn-16-OH contaminated with less than 1 % of its anti epimer. The syn-cyclopropylcarbinyl alcohol was somewhat sensitive and could not be obtained sufficiently pure, either by column chromatography or glpc, for an acceptable carbon-hydrogen analysis. The p-nitrobenzoate derivative was easier to purify and gave the expected combustion results. We recently discussed several criteria for assigning the stereochemistry at Cz for 2-bicyclo[6.1 .O]nonane derivative^.^" The same general principles appear to hold for syn- and anti-16-OH. Methylene addition to 15 with direction by the hydroxyl group gives the anti epimer,j and hydride attack at the least-hindered face of the carbonyl group reduces ketone 17 to syn-16OH. The more hindered hydroxyl group in syn-16-OH cannot form intermolecular hydrogen bonds as easily as anti-16-OH. The relative intensity of the 3630-cm-' ir band (free hydroxyl stretch) for syn-16-OH is significantly greater than for the corresponding band (3610 cm-l) of the anti epimer. Also, the neighboring cyclopropane ring shields the proton at Cz (3.26 ppm) in anti-16-OH while deshielding the corresponding proton (4.35 ppm) in syn-16-OH. The resulting chernicalshift difference is 1.09 ppm,6 which compares nicely with a calculated difference of 0.98-1.17 ~ p m . ~ " ( 5 ) We previously discussed the reasons for stereoselective methylene addition in these systems: C. D. Poulter, E. C. Friedrich, and S.Winstein, J , Amer. Chem. Soc., 91, 6892 (1969). (6) Models indicate that the preferred conformations of syn- and anti-2-bicyclo[6.1.O]nonanes and syn- and unti-2-bicyclo[7.1.0]decanes are similar.

15'

H anti-16-OH

trans-18-0H 91%

A trans, trans-19-0H 9%

were easily separated by chromatography on alumina impregnated with silver nitrate. The major isomer (trans-18-OH) comprised 91 % of the mixture. The homoallylic alcohol had an nmr spectrum similar to that for cis-18-OH and an ir spectrum with a strong band at 980 cm-l (vide supra). The affinity which trans-18-OH exhibited toward silver nitrate during chromatographyg is also characteristic of trans double bonds in medium rings. lo The minor isomer (9 of the mixture) exhibited ir and nmr spectra very similar to those reported for trans-bicyclo[5.2.0]nonan-trans-801.~~3'~On this basis, the minor isomer was assigned (7) The products assigned as hydrocarbons had short glpc retention times relative to syn-16-OH and cis-18-OH. Unfortunately, the hydrocarbon mixture was complex, and the structures of its components have not been determined. (8) A. T. Bloomquist, R. E. Burge, Jr., and A. C. Sucsy, J . Amer. Chem. SOC.,74,3636 (1952). (9) The easiest way to recover trans-18-OH was to stir the aluminasilver nitrate column packing with dilute ammonium hydroxide solution and extract the aqueous layer with ether. (10) A. C. Cope, K. Banholzer, H. Keller, B. A. Pawson, J. J. Whang, and H. J. S . Winkler, J . Amer. Chem. Soc., 87, 3644 (1965). (11) The nmr spectrum of truns,trans-19-OH is also similar to that of trons-bicyclo[4,2,0]octan-truns-7-ol, We wish to thank Professor Wiberg for a preprint of his full paper: K. B. Wiberg and J. G . Pfeiffer, ibid., 92, 553 (1970).

Poulter, Winsfein Homoallylic Ring Expansions and Contractions

4284 Table I. Isomerization Rate Constants in 80x DioxaneWater, 0.026 M HCIO, Alcohol

Temp, "C

104k,sec-1

syn-lbOH

25 50 75

anti-&OH

75

0.415 f 0.011 11.0 f 0 . 6 183a 5.99 f 0.18

Extrapolated from rates at 25 and 50"; AH* = 24.5 f 0.8 kcal/mol, AS* = 3.5 f 3 eu. 4

was very sensitive to acid. Its rate was measured at 25 and 50' and extrapolated to 75". The rate constants were calculated by comparing the relative areas of the cyclopropylcarbinyl alcohols with an internal standard as a function of time. The same results were obtained when the relative areas of products and internal standard were compared. Solvolysis studies of syn- and anti-16-OPNB were carried out using the sealed ampoule technique. 2,6-

Table II. Solvolysis Rate Constants, 80% Acetone-Water System Temp, "C 106k, sec-1 75.0 100.0 25.0 75.0 100.0 25.0

syn-lCOPNB cis-18-OBs anti-16-OPNB frans-l&OBs

8.61 f 0.14 14.5 f 3.7 0.384 f 0.009 0.368 f 0.010 4.39 fi 0.22 310 f 13

the structure of trans-bicyclo[6.2.0]decan-trans-9-o11z (trans,trans-19-0H). A careful glpc study showed that at least 0.3% of trans-18-OH and trans,trans-19-OH could have been detected in cis-18-OH and vice versa. Isomerizations were carried out with carefully purified samples of synand anti-16-OH which contained less than 0.1 % of the other epimer. We saw no crossover products between the syn- and anti-alcohols and conservatively estimate that less than 0.3% occurred. The progress of the isomerizations was followed by glpc. Samples were periodically withdrawn and quenched by vigorous shaking with anhydrous sodium carbonate. The rates of isomerization are listed in Table I. The syn epimer Scheme III h

IOPNB

AH*, kcal/mol

804'. acetone-water

H

21.6

AS*, eu

0.4

-15.5 f 1 . 2

24.9 f 0.5

-12.3 i 1.6

Lutidine was used to neutralize the p-nitrobenzoic acid generated during solvolysis. Large-scale solvolyses (ca. 1 g) were carried out in order to establish structures of the products by correlation of ir and nmr spectra and glpc retention times. After the components of the product mixture had been properly identified, product studies on dilute solutions (0.005-0.01 M ) were performed. Control experiments established that the products were stable to the reaction conditions. The results are summarized in Scheme 111. Rate constants were determined by titration of liberated p-nitrobenzoic acid and are listed in Table 11. The discrepancy between experimental and theoretical infinity titers and the amounts of homoallylic p-nitrobenzoates found in product studies were in excellent agreement. Product studies with cis- and trans-18-OBs were carried out using dilute solutions (0.005-0.010 M> with added 2,6-lutidine. Products were identified by coinjection with authentic samples on two different glpc columns. The results are summarized in Scheme IV. Scheme IV

syn-16-OPNB

80% acetone-water 2, Bhtidine

syn -16.OH

OH cL~-18-0H

26% OPNB

hydrocarbons 25%

&18-OPNB

45%

11%

OH

OH

~yn-16-OH

ci~18-OH

45%

30%

111 anti-16-0H 71%

an ti-16-OPNB

I

i-r

80% acetone-water 2,Ghtidine 1000

k

+

APNB

I

k

t

a

O

H

QH

acetone-water 2.6-lutidine 253

+

trans-18-OH

trans-18-OH 15%

trans-18-OPNB

13%

H trans, trans-19-0H 1%

(12) The trans,trans isomer is also the expected product from anti-16OH.**The nomenclature of the cyclobutyl ring system has previously been discussed.8*

Journal of the American Chemical Society

/

92:14

July 15, 1970

+ trans-18-OH

14%

A trans, trans-19-OH 1%

4285

The hydrocarbon products comprised a sizable portion of the products from syn-16-OPNB and cis-18-OBs. However, anti-16-OPNB and trans-18-OBs gave little, if any, hydrocarbon product. Thus, both brosylates solvolyzed with greater than 99.7 stereoselectivity, with regard to crossover between syn and anti products. The rate constant of cis-18-OBs was determined by titration of liberated acid in a conventional manner. The trans isomer was too reactive at 25 O (k = 3.10 X sec-1) for a normal titration. Known concentrations of base (freshly prepared in 80 acetone-water) were added to the solvent along with the indicator in separate test tubes. The tubes were allowed to equilibrate at 25". A standard amount of cis-18-OBs in 50 pl of dry acetone was added by syringe to each tube one at a time. The contents were well mixed, and the time required for a color change was measured with a stop watch. The infinity point was determined by conventional titration. l 3

Discussion The interconversions between syn-16-OH and cis18-OH and between anti-16-OH and trans-18-OH represent the first example of a full set of stereospecific homoallylic ring expansions and contractions. Solvolyses of syn-16-OPNB and cis-18-OBs gave similar product distributions, as did anti-16-OPNB and trans18-OBs. The ratio of cyclopropylcarbinyl to homoallylic product was slightly higher from cis-18-OBs and trans-18-OBs than from the corresponding cyclopropylcarbinyl p-nitrobenzoates. l 4 These results are only compatible with two noninterconverting, nonclassical homoallylic ions. Scheme V completes the fragmentary results found in other s y ~ t e m s . ~ The ~ - ~ reasons for expecting stereospecificity in these systems have been A few important condiscussed in various siderations form the basis of the explanation. Both the cyclopropane ring and the homoallylic double bond are involved in the transition states for ionization. The alkyl group at C g in syn- or anti-16 stabilizes the incipent homoallylic ion by utilization of the secondarysecondary CI-Cg bonding cyclopropane electrons in preference to the C1-Clo secondary-primary electrons. The exact magnitude of the stabilization is difficult to estimate since no products derived from the other possible, less-substituted homoallylic ion were observed. Schleyer and van Dine15 reported that the rate constant of cis-2-methyl-1-hydroxymethylcyclopropane 3,5-dinitrobenzoate is only 8.2 times that of cyclopropylcarbinyl 3,5-dinitrobenzoate, The small rate enhancement caused by the methyl group (and presumably other alkyl substituents) is not sufficiently large to explain the stereospecificity which we observe. l6 We can offer no explanation for this discrepancy other than the possibility that the rate constant reported for cyclopropylcarbinyl dinitrobenzoate (4.30 X lo-' sec- at 100O in 60 % acetone-water) represented predominant (13) A similar procedure has been used by Wiberg and Hess: K. B. Wiberg and B. A. Hess, Jr., J . Amer. Chem. Soc., 89,3015 (1967). (14) Similar results were found for syn-bicyclo[6.1 .O]non-2-yl p-nitrobenzoate and cis-cyclononen-4-yl p-bromobenzenesulfonate.aa Differences in temperature and (or) leaving groups are thought to be responsible for different product distributions from the same cation. (15) P. von R. Schleyer and G . W. van Dine, J . Amer. Chem. SOC.,88, 2321 (1966). (16) An additional alkyl substituent at the carbinyl carbon, as in syn- and anti-16-OPNB. could only diminish the effect of the alkyl substituent on the cyclopropane ring.

Scheme V

syn-16

cis-20

cis-18

anti-16

trans-20

cation

W

I

trans,tram.19-0H

acyl oxygen cleavage. Another feature essential to Scheme V is retention of stereochemistry between C1 and CPin the homoallylic ions. Both theoretical and experimental data3a suggest that the barrier to rotation about the c1-C~ bond is high. Stereospecific solvent attack on cis- and trans-20 along the reverse path for ionization gives the observed products. The stereospecificity found for [6.1.O]- and [7.1.O]cyclopropylcarbinyl systems does not extend to the lower homologs. Based on our current knowledge of medium rings, Scheme V would be expected to be marginal for the [6.1.0]system and fail for [5.1.0],[4.1.0], [3.1.0], and [2.1.0]systems. The strain energy of trans18 should increase rapidly as the size of the larger ring is decreased. For example, trans-cyclooctene is 8 kcal/ mol less stable than its cis isomer," and trans-cycloheptenonels is so reactive that it is stable only near liquid nitrogen temperature. The geometry at C1 and Cz in trans-20 should approach that of a trans double bond. We found that, whereas solvolyses of the [6.1.0] and [7.1.0] systems were stereospecific, those with smaller rings were Careful product studies on the smaller homologs (five-, six-, and seven-membered rings) are not complete, although there is some indica(17) A. C. Cope, P. T. Moore, and W. R. Moore, J. Amer. Chem. 1744 (1960). (18) (a) E. J. Corey, M. Tada, R. La Mahieu, and L. Libit, ibid., 87, 2051 (1965): (b) P. E. Eaton and K. Lin, ibid., 87, 2052 (1965).

Soc., 82,

Poulter, Winstein

Homoallylic Ring Expansions and Contractions

4286

tion that more than one nonclassical intermediate may be involved. Several possible explanations, such as equilibration between homoallylic and (or) symmetrical homoallylic cations, l9 cannot be ruled out without additional study. Also, the effect of ion pairing on the stereochemistry of the lower homologs is unknown.20 The seven-carbon bridges in cis- and trans-20 undoubtedly destabilize the parent isomeric cations. Models indicate that steric interactions are greater for trans-20 than its cis isomer. The ring strain of the trans cation is much greater when the larger ring is reduced to six carbon atoms. Yet, ionization of antibicyclo[6.1 .O]non-2-yl p-nitrobenzoate gives a transhomoallylic intermediate (>99.99 %).3a In the latter case, the driving force for backside assistance from the more substituted arm of the cyclopropane ring must be considerable. Solvolysis of trans-18-OBs proceeded about 1000 times faster than solvolysis of its cis isomer. The trans brosylate is probably somewhat more strained than c i s - 1 8 - O B ~ ;however, ~~ the strain of a trans orientation between CI and Cz is probably magnified at the transition state leading to trans-20. Thus, it seems unlikely that the strain associated with the trans double bond is responsible for the enhanced rate of trans-18OBs, since the increased ground-state strain of trans18-OBs should be accompanied by increased strain at the transition state leading to trans-20. The relative orientations between the double bonds and the leaving groups in cis- and trans-18-OBs seem to offer a rational explanation of the kinetic results. The double bond of trans-18-OBs is properly oriented to assist ionization of the brosylate, while several probable orientations for cis-18-OBs d o not appear to be as favorable for assisting ionization. As the transition state between trans-20 and trans-18-OH is approached from trans-20, the C1-C2 bond length should shorten considerably. The 7 carbon bridge is large enough to accommodate this development without inducing a large energy difference in the transition state leading to anti-16-OH and trans18-OH. Both isomers are found in the product mixture. When the bridge is reduced to 6 carbon atoms, ring strain at the transition state leading to trans homoallylic product apparently becomes so great that none is A small amount of cyclobutyl product (trans,trans19-OH) was found in the product mixture of anti-16OPNB. We feel that trans-20 isomerized stereospecifically to a nonclassical cyclobutyl ion which gave trans,trans-19-OH, in analogy with the isomerization of anti-bicyclo[6.1 .O]nonan-2-01 to trans-bicyclo[5.2.0]nonan-trans-8-01.~~In the latter case it was demonstrated that at least one cyclobutyl ion was involved, in addition to the trans-homoallylic ion. The decrease in cyclobutyl product as the larger bridge is expanded by one carbon atom probably results from a combination of factors. Part of trans-20 is consumed by sol(19) It must be remembered that resonance-stabilized cations in these systems are sufficiently stable to permit some equilibration before solvent collapse. Initial product studies with syn- and anri-16-OPNB in the presence of 0.02 M acetate gave up to 10% acetate products. The relative ratios of cyclopropylcarbinyl to homoallylic product were identical for alcohols and acetates. (20) It is obvioiis from the amount of internal return to homoallylic product that ion pairing is important in these systems. (21) frans-Cyclodecene is 2.2 kcal/mol less stable than cis-cyclodecene in acetic acid at 25 '. 17

Journal of the American Chemical Society j 92:14 j July 15, 1970

vent attack at Cs to give t r a n s - S O H . Also, trans30 is considerably less strained than its lower homolog, and the relief of strain energy by isomerization to a cyclobutyl cation is not as great. We would like to emphasize the synthetic utility of stereospecific homoallylic ring expansions for generating either cis or trans double bonds in medium rings. Starting with cis-cyclononen-3-01 (15) gram quantities of cis-18-OH were obtained in an overall yield of 59%, and gram quantities of trans-18-OH, in an overall yield of 47%. The only sizable losses were incurred during methylene addition to 15. Each of the homoallylic alcohols could be prepared with less than 1 % of the other without resorting to any tedious separations. Homoallylic ring expansions have been used extensively in this laboratory in synthetic sequences for preparation of possible precursors for the heptahomotropilium cation. In theory, trisubstituted double bonds could also be generated stereospecifically from the appropriate a-methylcyclopropylcarbinyl derivative. Experimental Section22 cis-Cyclononen-3-01 (15). cis-Cyclononen-3-01 was prepared by the procedure described by Santelli and coworkers,23 bp 70-73" (0.3 mm), lit.zs 62" (0.5 mm). anti-Bicyclo[7.l.0]decan-2-ol(anfi-16-OH).z4 Treatment of 12.0 g (0.086 mol) of 15 with 12.3 g (0.19 mol) of zinc-copper couple and 46.0 g (0.172 mol) of methylene iodide gave 8.66 g (66%) of a colorless, viscous liquid which crystallized in the condenser during distillation: mp 54.5-55.5'; ir (CCla) 3610, 3400, 3060, 2990, 2930,2860, 1440, 1060, 1025, 1005, 963, and 848 cm-l; nmr (CClr) 0.11 (1, m, endo H at GO),0.63 and 1.4 (15, m, H at CI, H at c3-C~ and exo H at Go), 2.28 (1, s, hydroxyl H), and 3.26 ppm (1, m, H at Cz). Anal. Calcd for C10H180: C, 77.87; H, 11.76. Found: C, 77.88; H, 11.66. A 1.992-g (0.013 mol) sample of anti-16-OH was allowed to react with 2.452 g (0.013 mol) of p-nitrobenzoyl chloride in 10 ml of anhydrous pyridine at -5" for 48 hr. Work-up and two recrystallizations from petroleum ether gave 3.401 g (87%) of a pale yellow solid: mp 123.5-124.5"; ir (CC14) 3110, 3060, 3000, 2920, 2850, 1715, 1605, 1525, 1343, 1270, 1110, 1095,940, and 870 cm-l; nmr (CC14)0.04 (1, m, endo H at Clo)0.5-2.1 (15, m, H at Ci, H at C3Cs. and exo H at Clo), 4.85 (1, m, H at CZ),and 8.05 ppm (4, s, aromatic H). Anal. Calcd for C I I H U N O ~ :C , 67.31; H, 6.98; N, 4.62. Found: C, 67.16; H,6.99; NI 4.36. Bicyclo[7.1.O]decan-2-one (17). A solution of 4.980 g (0.032 mol) of anti-16-OH in 75 ml of dry acetone was cooled in an icemethanol bath. To the cold solution was added 9.0 ml of Jones reagent.26 After 5 min, excess reagent was quenched with 5 ml of 2-propanol. The resulting green suspension was poured into 100 ml of water and the solid green residue was dissolved in an additional 100-ml portion of water, The combined aqueous solutions were extracted with petroleum ether. The combined extracts were washed with saturated sodium bicarbonate solution and dried over anhydrous magnesium sulfate. Solvent was removed at reduced pressure to give 4.76 g (96%) of a colorless liquid. Samples for spectra and combustion analysis were purified by glpc: ir (CC14) 3000, 2930, 2860, 1685, 1460, 1452, 1430, 1385, 1370, and 1097 cm-1; nmr (CCl4) 0.2-2.0 (1 5, m) and 2.6 ppm (1, m). Anal. Calcd for CloHlsO: C, 78.90; H, 10.59; Found: C, 79.02; H, 10.65. syn-Bicyclo[7.1.0]decan-2-ol (syn-16-OH). To a stirred suspension of 0.429 g (0.045 equiv) of lithium aluminum hydride in 50 ml of dry diethyl ether was added 4.65 g (0.031 mol) of 17. The (22) The general experimental techniques have been previously described. (23) M. Santelli, M. Bertrand, and M. Ronco, Bull. SOC.Chim. Fr., 3273 (1964). (24) A detailed procedure similar to that used to convert 15 to anfi16-OH has been de~cribed.~' (25) A. Bowers, T. G. Halsall, E. R. H. Jones, and A. J. Lemin, J . Chem. Soc., 2548 (1953).

4287 mixture was allowed to stir for 12 hr before excess hydride was carefully decomposed by dropwise addition of saturated ammonium chloride solution. The addition was continued until the inorganic material precipitated, leaving a colorless ethereal layer. The ether was decanted, and the precipitate was washed with additional portions of ether. The combined ether layers were dried. Solvent was removed at reduced pressure to give 4.30 g (91 %) of a colorless liquid: ir (CC14) 3630, 3480, 3060, 2990, 2920, 2860, 1510, 1470, 1440, 1078, 1025, 995, and 845 cm-l; nmr (ccl4) 0.12-1.15 (4, m, H at C1, Cg and GO),1.5 (12, m, H at CS-CS), 1.95 (1, s, hydroxyl H), and 4.35 ppm (1, m, H at CZ). The alcohol was quite sensitive and could be stored for long periods only at -15'. Repeated attempts to obtain an analytical sample which gave a combustion analysis within 0.3 % for both carbon and hydrogen failed. Anal. Calcd for CloHlsO: C, 77.87; H , 11.76. Found: C, 77.36; H, 11.87. A 2.012-g (0.013 mol) sample of syn-16-OH was allowed to react with 2.455 g (0.13 mol) of p-nitrobenzoyl chloride in 10 ml of dry pyridine at - 5 " for 72 hr. Work-up and recrystallization from petroleum ether gave 2.499 g (64%) of a pale yellow solid: mp 95-96'; ir (CC14) 3110, 3060, 2980, 2920, 2850, 1723, 1525, 1345, 1270, 1110, 1095, and 870 cm-1; nmr (CClr) 0.15 (1, m,endoHat Clo), 0.5-2.4 (15, m, H at C1, H at C3-C9, and exo H at G O ) ,5.70 (1, m, H at Cz),and 8.18 ppm (4, m, aromatic H). Anal. Calcd for C17Hz1N04: C, 67.31; H , 6.98; N , 4.62. Found: C,67.27; H,7.02; N,4.52. cis-Cyclodecen-4-01 (cis-18-OH). A mixture of 1.100 g (7.2 mmol) of syn-16-OH, 40 ml of dry dioxane, and 10 ml of 0.129 N perchloric acid was stirred at 75' for 1 hr. The solution was poured into 100 ml of water, and the resulting suspension was extracted with pentane, The combined pentane extracts were washed with sodium bicarbonate solution and dried. Solvent was removed at reduced pressure to give 0.990 g (90%) of a colorless oil which crystallized on standing. A 100-mgportion was recrystallized from petroleum ether to give white needles: mp 55.0-57.5'; ir (CS) 3620, 3360, 3000, 2980, 2940, 2920, 2900, 2860, 2850, 1650, 1470, 1450, 1440, 1010, 1000, 725, and 710 cm-1; nmr (CDCL) 1.46 (10, m, H at C3-C9), 2.3 (4, m, H at C3 and Clo),3.75 (1, s, hydroxyl H), 3.8 (1, m, H at C4),and 5.47 ppm (2, m, H at C1 and Cz). Anal. Calcd for CI0Hl80: C, 77.87; H , 11.76. Found: C, 77.86; H , 11.90. Following a general procedure for preparing reactive brosylates,& 126 mg (0.82 mmol) of cis-18-OH and 210 mg (0.82 mmol) of pbromobenzenesulfonyl chloride were allowed to stand at - 5" for 96 hr. Work-up of the mixture and recrystallization gave white plates (61%): mp 88-89' dec; ir (CSI) 3000, 2920, 2860, 2840, 1385, 1360, 1270, 1180, 1170, 1060, 1010, 950, 925, 910, 890, 815, 735, and 600 cm-1; nmr (CDC13) 1.4 (10, m, H at CS-CS),2.2 and 2.5 (4, m, H at C3 and Clo),4.75 (1, m, H at C4), 5.43 (2, m, H at Czand Ca),and 7.70 ppm (4, m, aromatic H). Anal. Calcd for Cl6Hz1SO3Br: C, 51.47; H , 5.67. Found: C, 51.68; H , 5.89. trans-Cyclodecen-4-01 (trans-18-OH). Following the procedure described for cis-lI)-OH, a mixture of 1.100 g (7.2 mmol) of anti16-OH in 40 ml of dioxane and 10 ml of 0.129 N perchloric acid was heated at 75" for 12 hr. Work-up of the mixture gave 0.992 g (90%) of a colorless oil, which was composed of two components (91 and 9 % by glpc). The mixture was chromatographed on alumina impregnated with 20% (by weight) of silver nitrate. The minor isomer (81 mg) was eluted with 1 :10 ether-pentane, and the major isomer was desorbed by stirring the column packing with 10% ammonium hydroxide solution. The aqueous layer was extracted with petroleum ether. The combined extracts were washed with water and dried. Solvent was removed at reduced pressure to yield 629 mg of a colorless oil. Analytical samples were collected by glpc: ir (CSz) 3620, 3340, 3020, 2920, 2840, 1660, 1450, 1435, 1045, 1010, and 980 cm-1; nmr (CC14) 1.38 (10, m, H at CSCS), 2.1 (4, m, H at CJ and G O ) ,3.7 (1, m, H a t C4), 4.35 (1, s, hydroxyl H), and 5.47 ppm (2, m, H at C1and Cz). Anal. Calcd for C10H180: C, 77.87; H, 11.76. Found: C, 77.65; H, 11.57. Following the procedure used to prepare cis-18-OBs, 159 mg (1.10 mmol) of trans-18-OH and 281 mg (1.10 mmol) of p-bromobenzenesulfonyl chloride were allowed to stand at -5" for 96 hr. Work-up gave 201 mg of a colorless, viscous oil which resisted numerous attempts to induce crystallization: ir (CSz) 3020, 2920, 2840, 1350, 1270, 1180, 1090, 1060, 1005, 975, and 890 cm-l; nmr (CDC13) 1.3 and 2.0-2.5 (14, m, H at C3 and Cs-Clo), 4.6 (1, m, H at C4), 5.4 (2, m, H at C1 and Cz), and 7.70 ppm (4, d, aromatic H).

rrans-Bicyclo[6.2.0]decan-trans-9-ol (trans,trans-19-OH). The minor component produced by acid-catalyzed isomerization of anti-18-OH was isolated by column chromatography. The compound (81 mg) was a colorless oil. Samples for analysis were purified by glpc: ir (ccl4) 3620, 3320, 2960, 2910, 2840, 1460, 1450, 1440, 1435, 1135, and 1065 cm-l; nmr (CClJ 1.1-2.3 (16, m, H at C1-Cs and Clo), 3.43 (1, m, H at Cg), and 4.27 ppm (1, s, hydroxyl H). Anal. Calcd for CloH1.0: C. 77.87:, H., 11.76. Found: C. .. .77.69; H , 11.75. Acid-Catalyzed Isomerization of syn-16-OH and anti-16-OH. The general procedure for kinetic and product determinations have been d e ~ c r i b e d . ~Glpc ~ analyses were carried out with 10 ft X lis in. Carbowax 20M or DEGS columns at 160" (Table 111). Table 111. Relative Glpc Retention Timesa Compd

Relative time

17 15 ~nti-16-OH rrans,rrans-19-OH syn-16-0H trans-18-OH cis-18-OH

1 .oo 1.36 1 . 50b 1.66 1.72 1.99 2.06 ~~

On 5 % Carbowax 20M, 160". proximately 13 min. a

At 160", elution took ap-

Control experiments established that as little as 0.3% of anti. 16-OH, trans-18-OH, and trans,rrans-19-OH could be seen in syn16-OH and cis-18-OH and aice aersa. In all cases no crossover products were detected. Product results are summarized in Scheme I1 and kinetic results are summarized in Table I. Kinetic Procedures. Except for trans-18-OBs the procedures have been described. 3a The trans-brosylate was too reactive at room temperature to permit conventional titration. An 80-pl portion of trans-18-OBs was dissolved in 320 pl of dry acetone, and the resulting mixture was kept in a tightly stoppered vial. A 40-p1 portion of the solution was added to 5.00 ml of 80% acetone-water and the resulting solution was placed in a 25" bath for 8 hr. One drop of a 1 % solution of bromothymol blue in acetone was added. The resulting mixture was titrated with a freshly prepared solution of sodium hydroxide in 80% acetone-water to determine the end point. A zero point with 5.00 ml of 80% acetone-water and 1 drop of indicator was also determined. Then portions of the base solution corresponding to 20-90% of the infinity titer in 10% steps were placed in eight test tubes. The volume was made up to 5.0 ml, and 1 drop of indicator was added. The tubes were stoppered and allowed to equilibrate at 25". One by one, 40-pl portions of trans-18-OBs in acetone were added to the solvent, base, and indicator solutions. The contents of the test tubes were well mixed, and the time required for a color change (blue to green) was measured with a stop watch. First-order rate constants were calculated in the usual manner. Preparative Solvolysis of syn-Bicyclo[7.l.O]dec-2-yl p-Nitrobenzoate (syn-16-OPNB). A solution of 1.ooO g (3.30 mmol) of syn-16-OPNB and 1.070 g (10.0 mmol) of 2,6-lutidine in 150 ml of 80% acetone-water was heated at 100" for 150 min (ca. 10halflives). The solution was poured into saturated sodium chloride solution, and the resulting suspension was extracted with pentane. The combined pentane extracts were washed with water and dried. Solvent was removed at reduced pressure to give 1.231 g of a light yellow oil. Chromatography on Woelm alumina, activity 11, gave, in order of elution, 23 mg of hydrocarbons, 139 mg of pale yellow solid, and 379 mg of a mixture of alcohols. The remainder of the sample was 2,6-lutidine. The alcohol mixture consisted of syneach was collected by glpc 16-OH (3973 and cis-18-OH (61 and identified by ir spectra. The pale yellow solid (cis-18-OPNB) was recrystallized from petroleum ether: mp 73-74"; ir ( C 8 ) 3000,2980,2920,2860,1720, 1600, 1340, 1270, 1110, 1095, and 715 cm-l; nmr ( C S ) 1.1-2.9 (14, m, H at Cs and Cs-Clo), 5.15 (1, m, H at C4), 5.35-5.75 (2, m. H at Cl and Cz),and 8.20 ppm (4, s, aromatic H). Anal. Calcd for C I ~ H Z ~ N OC,~ :67.31; H , 6.98; N, 4.62. Found: C, 67.37; H,6.85; N,4.64. Preparative Solvolysis of anti-Bicyclo[7.1.O]dec-2-yl p-Nitrobenzoate (anti-16-OPNB). Following the procedure described for syn-16-OPNB, a solution of 1.001 g (3.31 mmol) of anti-16-OPNB

Poulter, Winstein

z);

1 Homoallylic Ring Expansions and Contractions

4288

and 1.068 g (10.0 mmol) of 2,6-lutidine in 150 ml of 80% acetonewater was heated at 100" for 48 hr (ca. 10 half-lives). Work-up gave 1.319 g of a pale yellow oil. Column chromatography separated the residue into three fractions. The minor fraction (138 mg) was a pale yellow solid. The middle fraction (352 mg) was shown by glpc retention times and ir spectra to consist of anti-16OH (88 %), trans-18-OH(lZX), and truns,trans-19-OH (1 %). The remainder of the sample was 2,6-lutidine. Recrystallization of the minor fraction from petroleum ether (trans-18-OPNB) gave 124 mg of a pale yellow solid: mp 54-55"; ir (CSr) 3020,2920,2860, 1720, 1605, 1340, 1270, 1110, 1100, 1015, 980,870, and 720 cm-1; nmr (CDC13) 1.2-2.9 (14, m, H at C3 and C5-Clo),5.2 (1, m, H at C4),5.6 (2, m, H at CI and Cl), and 8.16 ppm (4, s, aromatic H).

Anal. Calcd for C1THz1NO.I: C, 67.31; H, 6.98; N, 4.62. Found: C, 67.22; H, 6.89; N,4.47. Analytical Product Solvolyses. Solutions 0.01 M in p-nitrobenzo-

ate or pbromobenzenesulfonate and 0.02 M in 2,6-lutidine were heated at the appropriate temperature for 10 half-lives. Acetonewater (80%) was used in all of the studies. Glpc analyses at 80 and 160" were carried out without prior work-up. The products were identified by glpc retention times, and the results are summarized in Schemes 111 and IV. Mixtures of 1 equiv of the product alcohols, 1 equiv of p-nitrobenzoic acid, and 2 equiv of 2,6-lutidine were heated for the same time periods required for solvolysis of the precursor p-nitrobenzoate. Comparison of glpc traces before and after heating established that the products were stable to the reaction conditions.

Studies in Mass Spectrometry. XIII.' Stereospecific Electron Impact Induced Fragmentation Processes in Some Tricyclic Diesters Joseph Deutsch2 and Asher Mandelbaum

Contribution from the Department of Chemistry, Israel Institute of Technology, Haifa, Israel. Received December 1, 1969 Abstract: endo-Dimethyl esters l a , 2a, 3a, and 4a undergo elimination of methanol under electron impact through a seven-centered transition state. Stereoisomeric exo diesters lb, 2b, 3b, and 4b do not appreciably eliminate methanol. Loss of methoxyl radical from the molecular ion has been observed instead. trans-Diesters IC,2c, 3c, and 4c eliminate methanol under electron impact through a five-centered transition state. An additional stereospecific hydrogen migration accompanying retro-Diels-Alder fragmentation has been observed in the endo isomers.

T

he effect of stereochemical factors on the behavior of organic ions produced in the ion source of a mass spectrometer has been of considerable interest since mass spectrometry became a widely used tool in organic structural analysis. Differences in the abundance of certain ions obtained from stereoisomers under electron impact (or photoionization) have been observed in many cases, and attempts have been made to relate these differences to distinctive features of the parent compounds. One possible argument for the difference in the abundance of certain fragment ions originating from different stereoisomeric parent ions is the existence of sterically controlled fragmentation processes which may be preferred in one of the isomers. This is usually the explanation for the difference in the abundance of the products of rearrangement processes which are believed to occur via cyclic transition states.4 Stereoisomers in which such cyclic transition states are possible without the occurrence of prior skeletal rearrangements can be expected to give rise to more abundant rearrangement ions than those isomers, in (I) Part XII: M. Solomon and A. Mandelbaum, Chem. Commun., 890 (1969). (2) Based in part on the D.Sc. and M.Sc. Theses of J. Deutsch. (3) K. Biemann, "Mass Spectrometry, Organic Chemical Applications," McGraw-Hill Book Co., Inc., New York, N. Y., 1962, p 144; S. Meyerson and A. W. Weitkamp, Org. Mass. Spectrom., 1,659 (1968), and references cited therein; M. M. Green, R. J. Cook, J. M. Schwab, and R.B. Roy, J . Amer. Chem. Soc., 92,3076 (1970). (4) F. W. McLafferty in "Mass Spectrometry of Organic Ions," F. W. McLafferty, Ed., Academic Press, New York, N. Y., 1963, p 331 ; R. G. Cooks, Org. Muss Specfrom., 2, 481 (1969), and references cited therein.

which a different group may migrate or skeletal rearrangements must be postulated to enable attainment of a cyclic transition state. Relatively few cases have been reported in which radical difference exists between the mass spectra of stereoisomers. I n the course of our study of the specific double hydrogen migration in the adducts of bi-l-cycloalken1-yls and p-benzoquinones we examined the mass spectra of the three isomeric methyl esters of endo-, exo-, and trans-1,2,3,3a,4,5,5a,6,7,8-decahydroindacen4,5-dicarboxylic acids, l a , lb, and 1c6 (Figure 1). The mass spectra differed very much in the high mass range in the fragmentation of the carbomethoxy group. While in the case of the exo isomer l b methoxyl radical C H 3 0 was lost from the molecular ion (m/e 247 ion alb, abundance ratio alb/M+ = 0.81), methanol elimination gave rise to very abundant ions in the case of the endo (m/e 246 ion bl,, abundance ratio bl,/M+ = 2.1) and trans (m/e 246 ion b,,, abundance ratio bl,/M+ = 2.3) isomers l a and IC (see Figure 1). Elimination of methanol gave rise to a very low peak in the mass spectrum of the exo isomer lb. The same was true for the loss of a methoxyl radical from the molecular ion of l a and IC(see Figure 1). This different behavior under electron impact was found to be general for the three isomeric forms of homologous diesters 1-4 having various rings A and C. All exo isomers (series b) exhibit M - 31 ions (M - 32 ( 5 ) J. Deutsch and A. Mandelbaum, J . Amer. Chem. SOC.,91, 4809 (1969). (6) H. Christol and M. Levy, Bull. SOC.Chim. Fr., 3046 (1964).

Journal of the American Chemical Society / 92:14 1 July 15, 1970